happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

vulkan: add RTE variants of exp shader (#16165) 

This fixes some failures on Turing where "round to zero" rounds to the max f16
value but the CPU reference value is infinite.
This commit is contained in:
Jeff Bolz 2025-09-22 00:37:17 -05:00 committed by GitHub
parent 4d0a7cbc61
commit a20d810d79
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 17 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
ggml/src/ggml-vulkan/ggml-vulkan.cpp
View File

@ -3391,7 +3391,6 @@ static void ggml_vk_load_shaders(vk_device& device) {
ggml_vk_create_pipeline(device, device->pipeline_ ## name [0], #name "_f32", name ## _f32_len, name ## _f32_data, "main", 2, sizeof(vk_op_push_constants), {512, 1, 1}, {}, 1); \ ggml_vk_create_pipeline(device, device->pipeline_ ## name [0], #name "_f32", name ## _f32_len, name ## _f32_data, "main", 2, sizeof(vk_op_push_constants), {512, 1, 1}, {}, 1); \
ggml_vk_create_pipeline(device, device->pipeline_ ## name [1], #name "_f16", name ## _f16_len, name ## _f16_data, "main", 2, sizeof(vk_op_push_constants), {512, 1, 1}, {}, 1); ggml_vk_create_pipeline(device, device->pipeline_ ## name [1], #name "_f16", name ## _f16_len, name ## _f16_data, "main", 2, sizeof(vk_op_push_constants), {512, 1, 1}, {}, 1);
CREATE_UNARY(exp)
CREATE_UNARY(gelu) CREATE_UNARY(gelu)
CREATE_UNARY(gelu_erf) CREATE_UNARY(gelu_erf)
CREATE_UNARY(gelu_quick) CREATE_UNARY(gelu_quick)
@ -3403,6 +3402,17 @@ static void ggml_vk_load_shaders(vk_device& device) {
CREATE_UNARY(hardswish) CREATE_UNARY(hardswish)
#undef CREATE_UNARY #undef CREATE_UNARY
#define CREATE_UNARY_RTE(name) \
if (device->float_controls_rte_fp16) { \
ggml_vk_create_pipeline(device, device->pipeline_ ## name [0], #name "_f32_rte", name ## _f32_rte_len, name ## _f32_rte_data, "main", 2, sizeof(vk_op_push_constants), {512, 1, 1}, {}, 1); \
ggml_vk_create_pipeline(device, device->pipeline_ ## name [1], #name "_f16_rte", name ## _f16_rte_len, name ## _f16_rte_data, "main", 2, sizeof(vk_op_push_constants), {512, 1, 1}, {}, 1); \
} else { \
ggml_vk_create_pipeline(device, device->pipeline_ ## name [0], #name "_f32", name ## _f32_len, name ## _f32_data, "main", 2, sizeof(vk_op_push_constants), {512, 1, 1}, {}, 1); \
ggml_vk_create_pipeline(device, device->pipeline_ ## name [1], #name "_f16", name ## _f16_len, name ## _f16_data, "main", 2, sizeof(vk_op_push_constants), {512, 1, 1}, {}, 1); \
}
CREATE_UNARY_RTE(exp)
#undef CREATE_UNARY_RTE
#define CREATE_GLU(name) \ #define CREATE_GLU(name) \
if (device->float_controls_rte_fp16) { \ if (device->float_controls_rte_fp16) { \
ggml_vk_create_pipeline(device, device->pipeline_ ## name [0], #name "_f32_rte", name ## _f32_rte_len, name ## _f32_rte_data, "main", 3, sizeof(vk_op_glu_push_constants), {512, 1, 1}, {}, 1, true); \ ggml_vk_create_pipeline(device, device->pipeline_ ## name [0], #name "_f32_rte", name ## _f32_rte_len, name ## _f32_rte_data, "main", 3, sizeof(vk_op_glu_push_constants), {512, 1, 1}, {}, 1, true); \

1
ggml/src/ggml-vulkan/vulkan-shaders/exp.comp
View File

@ -1,5 +1,6 @@
#version 450 #version 450
#include "rte.comp"
#include "generic_head.comp" #include "generic_head.comp"
#include "types.comp" #include "types.comp"

7
ggml/src/ggml-vulkan/vulkan-shaders/vulkan-shaders-gen.cpp
View File

@ -704,8 +704,11 @@ void process_shaders() {
string_to_spv("upscale_f32", "upscale.comp", {{"A_TYPE", "float"}, {"B_TYPE", "float"}, {"D_TYPE", "float"}}); string_to_spv("upscale_f32", "upscale.comp", {{"A_TYPE", "float"}, {"B_TYPE", "float"}, {"D_TYPE", "float"}});
string_to_spv("exp_f16", "exp.comp", {{"A_TYPE", "float16_t"}, {"D_TYPE", "float16_t"}}); for (auto rte : {false, true}) {
string_to_spv("exp_f32", "exp.comp", {{"A_TYPE", "float"}, {"D_TYPE", "float"}}); std::string suffix = rte ? "_rte" : "";
string_to_spv("exp_f16" + suffix, "exp.comp", {{"A_TYPE", "float16_t"}, {"D_TYPE", "float16_t"}, {"RTE16", rte ? "1" : "0"}});
string_to_spv("exp_f32" + suffix, "exp.comp", {{"A_TYPE", "float"}, {"D_TYPE", "float"} , {"RTE16", rte ? "1" : "0"}});
}
string_to_spv("gelu_f16", "gelu.comp", {{"A_TYPE", "float16_t"}, {"D_TYPE", "float16_t"}}); string_to_spv("gelu_f16", "gelu.comp", {{"A_TYPE", "float16_t"}, {"D_TYPE", "float16_t"}});
string_to_spv("gelu_f32", "gelu.comp", {{"A_TYPE", "float"}, {"D_TYPE", "float"}}); string_to_spv("gelu_f32", "gelu.comp", {{"A_TYPE", "float"}, {"D_TYPE", "float"}});
string_to_spv("gelu_erf_f16", "gelu_erf.comp", {{"A_TYPE", "float16_t"}, {"D_TYPE", "float16_t"}}); string_to_spv("gelu_erf_f16", "gelu_erf.comp", {{"A_TYPE", "float16_t"}, {"D_TYPE", "float16_t"}});